Microsoft запустила первую AI-модель для кибербезопасности и agentic-платформу

Microsoft, in the week of July 27, 2026, expanded its lineup of AI tools for cybersecurity: the company unveiled its first proprietary AI-based security model and a new agentic platform where AI agents carry out defensive tasks autonomously. TechCrunch reports.

What Microsoft launched

Microsoft released two products at once: the first AI model in its history trained specifically for cybersecurity tasks, and a platform that deploys autonomous AI agents on top of this model. The key difference from previous releases is that the company is not offering a separate assistant feature within an existing product, but a complete defense system where decisions are managed by AI.

• Announcement — week of July 27, 2026
• Microsoft's first AI model tailored specifically for cybersecurity
• New agentic platform: AI agents run defensive operations without constant human involvement
• Both products are part of the company's expanding security lineup
• News source — TechCrunch

How an agentic system differs from an assistant

Unlike a chat assistant, an agentic platform doesn't wait for a separate command at every step: the agent investigates an incident on its own, correlates signals from different sources, and either proposes or carries out response actions. For security operations center (SOC) teams, this is a shift from an "AI suggests to the analyst" model to an "AI acts on its own, while the human oversees" model.

Microsoft is building this logic into a dedicated security model rather than layering it on top of a general-purpose language model. The specialized model is trained on defense-specific data — logs, telemetry, threat signals — and therefore, by the company's design, interprets attack context more accurately than a general-purpose model. This approach reflects a broader trend: instead of one large universal model, vendors are increasingly releasing narrow domain-specific models for particular tasks — programming, law, medicine, and now security.

"Microsoft boosted its AI offerings in cybersecurity with the launch of its first security model and a new platform,"

TechCrunch reports.

Why this matters for the market

The launch of a specialized security model by one of the largest enterprise software vendors sets a direction for the entire industry. Microsoft has been developing its Security Copilot product for several years; its own security model and the agentic platform built on top of it are the next logical step, shifting part of the workload from analysts to autonomous agents.

The demand stems from a chronic shortage of security specialists and the speed of modern attacks, where only minutes remain to react. Autonomous agents promise to close the gap between the volume of threats and the number of people capable of handling them. For competitors — Google, CrowdStrike, Palo Alto Networks — Microsoft's move signals a need to accelerate their own agentic developments in defense.

Cybersecurity is a convenient first niche for autonomous agents: tasks here are repetitive and well formalized (alert triage, log correlation, initial incident sorting), and the outcome is measurable — response time and the number of threats closed. That's exactly why agentic AI is arriving in the SOC earlier than in many other corporate processes.

What remains unknown

In its brief announcement, Microsoft did not disclose key details about the new products: there is no information on pricing, the timeline for general availability, supported languages, or results on public security benchmarks. It is also unknown whether the new model will be available separately or only bundled with existing Microsoft Defender and Security Copilot products. Official Microsoft documentation remains the primary source for verifying specifications.

What this means

Cybersecurity is becoming one of the first mass-market arenas for agentic AI. Microsoft is betting on autonomous agents in a field where human analysts are in short supply and where minutes decide the outcome of an attack — setting a bar the rest of the market will have to answer to. For enterprise customers, this means that in the coming months, an increasing share of routine SOC work will shift to AI agents.